Soil Dynamics and Earthquake Engineering 158 (2022) 107280

Contents lists available at ScienceDirect

Soil Dynamics and Earthquake Engineering

journal homepage: www.elsevier.com/locate/soildyn

Harbor Bay Business Park liquefaction during Loma Prieta earthquake - A

critical state perspective
Jorge Macedo a, *, Luis Vergaray a, Curtis Jensen b, Renzo Cornejo a, Michael Jefferies c
a
Georgia Institute of Technology, Atlanta, GA, USA
b
Consulting Engineer, USA
c
Formerly Gulf Canada Resources, Canada

A R T I C L E I N F O A B S T R A C T

Keywords: Part of the Harbor Bay Business Park, Alameda in California, showed evidence of liquefaction during the 1989
Earthquake-induced liquefaction Loma Prieta earthquake, but this was accompanied by only small settlements and no significant damage to two
Critical state soil mechanics recently constructed office buildings. Digital CPT records are available from a reasonably comprehensive post-
State inversion
earthquake investigation. These, in combination with the large body of test data associated with the Park’s
Loma Prieta earthquake
development, have been used to assess the soil behavior at this site further. Using analysis based on critical state
soil mechanics, it appears likely that the presence of underlying Bay Mud was an important factor, with the
limited undrained strength of that stratum limiting the transmitted shear stresses to the overlying and liquefiable
sand fill but without preventing its liquefaction. This inferred behavior is consistent with a system (i.e., favoring
interaction between layers) response, which was also observed during the 2010–2011 Canterbury earthquake
sequence.

1. Introduction
This case history is within the Harbor Bay Isle project, which consists
of 940 acres of hydraulic fill dredged from San Francisco Bay and
pumped into the site after constructing perimeter dikes, starting late in
1966. Harbor Bay Isle is adjacent to the south, west, and north sides of
Bay Farm Island, a district of Alameda city, California. Both Harbor Bay
Isle and Bay Farm Island are separated from the rest of the city on
Alameda Island by the San Leandro channel. About 600 acres were
zoned as residential and the remainder as commercial; the latter is now
known as the Harbor Bay Business Park (HBBP). The HBBP borders on
the Oakland International Airport and the San Francisco Bay.
There were instances of liquefaction and non-liquefaction within
Harbor Bay Isle during the magnitude 6.9, 1989 Loma Prieta earthquake
that affected California. Liquefaction in the Harbor Bay Business Park
has been referred to as Site 4 in Mitchell et al. [1] and as Alameda Bay
Farm in Moss et al. [2]. However, these prior investigations used limited
cone penetration test (CPT) data and also, as conventional at the time,
regarded the hydraulic fill as the only soil of interest. Here we consider
in more detail a 0.12 km2 part of the HBBP, which was originally called
Plaza 6 and then renamed the International Teleport Plaza (ITP). Our
approach follows a critical state framework for both the dredged sand
fill and the underlying Bay Mud, with conventional reliance on surficial
manifestations of liquefaction (there were no strong motion or
piezometer records at this site).
The ITP area is enclosed by a South Loop Road and the Harbor Bay
Parkway, with an aerial view and site plan shown in Fig. 1. Fig. 2 il­
lustrates one of the two buildings constructed in the ITP before the 1989
Loma Prieta earthquake, corresponding to a standard low-rise com­
mercial development. The extent of liquefaction shown in Fig. 1 is from
an inspection of the time (Hallenbeck & Associates [3]) and the lique­
faction details are discussed in the manuscript; the buildings shown on
the photograph to the west and northwest of the ITP had not been
constructed in 1989, and liquefaction there was a greenfield condition.
2. Methodology
The HBBP case history is developed using critical state soil me­
chanics, and it is helpful to summarize this framework as it differs from
the current predominant practice. For sands, the current practice largely
uses the so-called NCEER method for assessing liquefaction triggering (e.
g., Youd et al. [4]) where fines content (measured or typically inferred as
in the case of the Cone Penetration Test – CPT) is used in addition to
penetration resistance. If sufficiently plastic, fine-grained soils are

* Corresponding author.
E-mail address: jorge.macedo@ce.gatech.edu (J. Macedo).

https://doi.org/10.1016/j.soildyn.2022.107280
Received 22 August 2021; Received in revised form 1 February 2022; Accepted 25 March 2022
Available online 27 April 2022
0267-7261/© 2022 Elsevier Ltd. All rights reserved.
J. Macedo et al.
Fig. 1. International Telegraph Plaza layout. Representative subsurface infor­
mation is shown in Fig. 3, a) Aerial view of the HBBP showing the ITP location
where liquefaction occurred during the 1989 Loma Prieta earthquake, b) Plan
of development from ~1985 showing site investigations and pro­
posed buildings.
Fig. 2. Building at 1420 harbor Bay Parkway.
characterized as ‘clay-like’ and taken as non-liquefiable. In contrast,
while still relying on the CPT as a key input, a critical state approach
incorporates more explicitly the soil’s intrinsic frictional strength and
compressibility in evaluating CPT data for liquefaction assessments.
The critical state approach to soil behavior originated with the U.S.
Army Corps of Engineers (USACE) nearly a century ago when con­
structing one of the world’s first liquefaction-resistant dams at Franklin
2
Soil Dynamics and Earthquake Engineering 158 (2022) 107280
Falls [5]. The engineering of this dam was based on the recognition that,
as soil comprises particles, the distortion (“shearing”) of soil generally
involves a change in the void ratio (e) as the particles move past each
other. If the soil is loose, the particles will tend to move to a denser state,
while the opposite will occur if the soil is dense. Two particular un­
derstandings then developed: i) void ratio evolved to a unique
critical-state value with distortion regardless of its initial value (Casa­
grande [6,7]), with this large-distortion value defining the critical state
line (CSL) and being the key to understanding soil behavior as any soil
that was looser would be contractive and so liquefiable if saturated
(hence the name ‘critical void ratio’); and, ii) soil strength developed
from the combination of ‘friction’ (i.e., the critical state friction) and
‘interlocking’ (i.e., dilatancy caused by distortion) which was formalized
by Refs. [8,9]. The missing idea at that time was how to link (i) and (ii);
this link was provided some forty years later and was that the limiting
dilatancy depended on the difference between the soil’s current void
ratio and its critical void ratio at the current mean effective stress, which
was defined as the state parameter (ψ ) by Been and Jefferies [10]. These
three ideas are key for a modern understanding of soil behavior and
apply regardless of a soil’s gradation, fines content, etc. (Jefferies [11]).
In particular, this framework was found necessary to understand recent
liquefaction failures of three large dams: Fundão [12]; Cadia [13]; and,
Brumadinho [14,15]. The critical state approach has also been found
helpful in unifying cyclic liquefaction data (e.g., Refs. [16–20]).
In applying the critical state approach to the HBBP case history, the
fines content’s notion is replaced by the soil’s stress ratio at critical state
- also known as soil’s intrinsic friction-i.e., Mtc = q/p, with q the
deviatoric stress and p the mean effective stress at critical state), its
compressibility (λ10 ), and additional mechanical properties (discussed
in section 4). Of note, under a critical state framework, a ‘sand like’ or
‘clay like’ separation is not required; different soils simply exhibit
different mechanical properties such as Mtc , λ10 . Correspondingly, one of
the key aspects of the case history discussed in this study becomes
assessing the properties of the soils involved. A further difference to the
NCEER methodology is that in a critical state approach, CPT data is not
reduced to a reference stress level, but instead, the data is used as
dimensionless stress-ratios in a formal inversion following the laws of
mechanics (the inversion process is discussed in section 5). Thus, the
final output is the site response being characterized in terms of the in-
situ state parameter, which is linked with mechanical soil properties.
These aspects are illustrated later in the manuscript.
3. Ground conditions
3.1. Regional setting
The field investigations at the ITP site extended through an existing
Fill, the Holocene Bay Mud sediments (HBM, often just called ‘Bay Mud’
in the documents of the time), the Merritt Sand, and at some instances
into the Pleistocene Bay Mud (PBM) sediments, also known locally as the
Old Bay Clay or just the Old Bay Mud. These formations are described in
detail in the subsequent sections.
This sequence is underlain by the Alameda formation and then a
dense Jurassic-Cretaceous bedrock of the Franciscan formation bedrock;
none of these strata were investigated at the ITP. Rogers and Figuers
[21] summarized data from some 200 deep borings in the greater
Oakland area, reporting a bedrock depression beneath Bay Farm Island
and Oakland International Airport with bedrock at a depth of about 300
m.
3.2. Site investigations
Initial geotechnical investigations of the HBBP began about 1979 and
included mechanical CPT’s and rotary test borings. As the site devel­
oped, further site investigations were carried out with the electronic CPT
used from the early 1980s for which digital records exist. As can be seen
J. Macedo et al.
in Fig. 1, the investigations amount to a grid of CPTs supplemented by
sampled borings (also illustrated in Fig. 1). Additional CPT data to ITP is
available from other investigations nearby, including digital data for
pre-earthquake conditions.
Seven CPTu’s were performed in the Fall of 1989 on sites where
liquefaction was observed, with the location of seven of these being
shown in Fig. 1 (denoted as ITP No1, etc.); these were modern CPTs with
the standard pore pressure sensor location (i.e., the u2 location). This
data is in digital form and is processed here using modern methods for
evaluating ground conditions from the CPT. These CPT soundings were
supplemented by a further twelve CPT soundings carried out in three
campaigns of four soundings, each that were separated by several weeks
to evaluate post-EQ aging effects. Shear wave velocities were also
measured at the site. CPT soundings have continued since 1990 as the
HBBP has been developed. For instance, CPT soundings were made in
the ITP in 2018 for a new development (Langan [22]), with some of
these to a depth of about 25 m below ground surface, deeper than what
was attained by the immediate post-earthquake CPTs of 1989. The
extensive CPT soundings (Langan [22]) have been complemented by
sampled borings, with some eighteen borings put down at the site to
define the fill and underlying natural ground; these borings extended to
the base of the Bay Mud.
Laboratory testing was limited to index and gradation during the
development of the HBBP (1980–2018). In 2020, a bulk sample of the
hydraulic fill was recovered from the ground surface in an area that had
been exposed by construction activities. The sampling location was
slightly more than 30 m north of 1420 Harbor Bay Parkway (one of the
two buildings that were on the ITP before the Loma Prieta earthquake)
within a current building construction site. This bulk sample has been
used for measuring the mechanical properties of the hydraulically
placed sands, as discussed later.
Fig. 3. Stratigraphy at HBBP from a deep CPT record (data courtesy of Langan).
3
Soil Dynamics and Earthquake Engineering 158 (2022) 107280
3.3. Stratigraphy
The first soil layers at the ITP consist of a hydraulic fill, which was
ubiquitously dense to very dense near the surface, grading from loose to
very loose at depth. The fill overlies natural soil, with the top of the HBM
layer defining the transition between fill and natural soil. The HBM is
underlain by Merritt Sand, underlain in turn by soils of the PBM
sequence. The water table lies about 1.5–2.5 m below the ground sur­
face, which is slightly above mean sea level. The CPT data indicates
hydrostatic conditions in the sands at the site. Fig. 3 shows the typical
ground conditions encountered within the HBBP to up to 25 m depth.
Stratigraphic boundaries have been picked using the combination of
CPT resistance (e.g., the tip resistance, qc ), the normalized measured
excess pore pressure (Bq ), and friction ratio (F), which is a standard
approach. The typical stratigraphic sequence at the ITP is illustrated in
Fig. 3. This sequence is also consistent with the stratigraphy at the Naval
Air Station (USGS [23]), which is at an approximated distance of 7 km
and it is the site of the nearest strong-motion station. Nine soil zones are
recognized within the upper 25 m of the ITP. Some stratigraphic
boundaries are sharp, for example, that of the HBM to Merritt Sand,
while those within the PBM are more gradational. Some variation in soil
behavior occurs within each identified stratum, which is usual. The
stratigraphic units are now discussed in turn using the typical profile in
Fig. 3.
Layers 1 and 2 are hydraulic fill, with Layer 1 being sufficiently
above the groundwater such that construction equipment could move on
it (possibly densifying it) while Layer 2 reflects the hydraulic fill in its as-
placed condition. The deeper parts of the hydraulic fill were interbedded
with soft clay trapped in the fill during filling and referred to as “slurry”
in logs and documents of the time. Numerous samples were tested for
their water contents. Samples with water contents of about 20% are
likely silty sands broadly similar to the gradation envelope for the fill:
Layer 3. Water contents in the 30%–40% range probably included sand/
J. Macedo et al. Soil Dynamics and Earthquake Engineering 158 (2022) 107280

clay mixtures, the clay being plastic: Layer 4. There were a few water
contents of 50% and above in layers logged mainly as sandy, but these
are believed to be isolated pockets. The boring logs also indicate that the
soft, plastic clay “slurry” was also encountered as thin seams as well as
thicker layers.
The boundary between the fill (about 20 years old in 1989) and the
underlying natural soils is the top of Layer 5, the Holocene Bay Mud. The
HBM was encountered in most of the borings, except for those near
Harbor Bay Parkway. The transition elevation did not vary much,
ranging from about 26.2 to 27.1 m and averaging 26.7 m with a standard
deviation of 0.4 m. This stratum thickens, but not smoothly, from
northwest to southeast across the ITP.
The top of layer 6 is the transition to much older soils and soils found
elsewhere in the region. Layer 6 is a silty sand, called Merritt Sand by
other workers (e.g., USGS [23]). It was encountered in all CPTs. The top
of layer 7 is the start of the PBM sequence. There are multiple sub-units
within the underlying Pleistocene Bay Mud identified as Layers 7, 8, and
9. These layers comprise clay and silt strata to a considerable depth; the
full depth of this unit was not identified in the HBBP, but it has a
thickness of some 40 m at the Naval Air Station.

4. Soil properties

4.1. Sand fill (layers 1 and 2)
The hydraulically placed sand fill has a median grain size range of
160 <D50 < 220 μm and a fines content of less than 20% (typically about
10%), illustrated in Fig. 4. A sample of this fill collected in 2018 was
used for triaxial testing to determine the mechanical properties,
following the protocols given in Jefferies & Been [24]. The CSL deter­
mined is shown in Fig. 5.
The triaxial testing on HBBP fill was supplemented using data on
similar sands, in particular, to quantify the effect of changing fines
content on the fill mechanical behavior. A large body of fill testing was
carried out for oil exploration in the U.S and Canadian arctic offshore,
deriving on data on the public domain (Jefferies and Been [25]). This
data was searched, and Isserk sand, a predominantly quartzitic sand
with no plastic fines as the collected fill, was identified as gradationally
similar to the HBBP fills (see Fig. 4). Interestingly, the CSL of Isserk sand
with 5% fine contents (indicated as Isserk 220/5 in Fig. 5) is also
consistent with that of the collected fill (see Fig. 5), which also has
similar fine contents. The tests on Isserk sand with 5% and 10% fine
contents are used in this study to include gradational effects on assessing
the in-situ state of the fill sands. For instance, the CSLs of the two Isserk
sand gradations are shown in Fig. 5 to illustrate the effect of changing
fines; increasing fines increases the slope of the CSL (λ10 ), a behavior
Fig. 5. Critical state locus of the HBBP fill sand and the Isserk sands with
gradations of 5% and 10%.
encountered with other sands (Jefferies and Been [25]). This feature will
be used subsequently on identifying the in-situ state of the fill sands (see
section 5).
The corresponding strength data representative of the fill materials is
shown in Fig. 6, with the strength properties Mtc (already defined) and
N, which is the volumetric coupling in a strength-dilatancy relationship
(q/p)max = Mtc − (1 − N) Dmin , where Dmin is the maximum dilatancy.
The effect of soil state on the limiting (“peak”) dilation rate is shown in
Fig. 7 through the state-dilatancy parameter X, which represents the
scaling of Dmin and the soil state, using the state parameter (ψ ) as defined
in Been and Jefferies [10]. Mtc , N, and X can be calculated from triaxial
tests with procedures detailed in Jefferies and Been [25], which we have

Fig. 6. Effect of the maximum dilatancy (Dmin ) on the drained strength of the
Fig. 4. Gradation envelope of HBBP fill, recovered HBBP sand, Isserk sand, and sand fill. ηmax corresponds to the maximum η during each test, and Mtc is the
examples of fill with slurry inclusion. stress ratio at the critical state.

4
J. Macedo et al. Soil Dynamics and Earthquake Engineering 158 (2022) 107280

Fig. 7. Control of maximum dilatancy (Dmin ) by state parameter (ψ ).
followed in this study. Sand properties are summarized in Table 1, with
the meaning of the properties annotated on the respective figures.
4.2. Compromised fill (Interbedded layers 3 and 4)
Layers 3 and 4 appear as intermingled soils within the extensive
Layer 2, with questionable lateral continuity. There are no triaxial tests
or similar on either 3 or 4 layers, so the properties must be estimated. In
the case of Unit 3, its compressibility is assessed (see section 5) from the
CPT data based on its ‘soil behavior type’. Unit 3 appears to retain the
characteristics of the parent sand fill but with increased compressibility
Fig. 8. Insitu void ratio of Holocene Bay Mud. The CSL and ψ contours are
approximated to match the observations in Figs. 18 and 19 and are shown only
for illustrative purposes. The CSL is estimated by recovering a ψ = − 0.05
(Fig. 19) for an initial void ratio of 2.04 at a mean pressure of 100 kPa (Fig. 18).
have found similar normally consolidated conditions in the HBM else­
where (e.g., Refs. [26–28]). These conditions are reflected in the void
ratios of the in-situ HBM samples being predominantly looser (wetter)
than the estimated critical state locus (see below), as shown in Fig. 8.
The initial compression index (Cc ) was largely in the range 1.05 < Cc <
1.37 with a representative Cc ~ 1.10. However, most unusually, the
swelling (elastic) index was small with a consistent Cs /Cc ~0.1; this is
about half the elastic void ratio recovery normally expected for soft clay.
The strength tests were only performed on samples that did not
display visible signs of disturbance. Strength was taken as the peak of the
stress-strain curve or the stress at 10% strain, whichever occurred first.
The data is shown in Fig. 9, normalized by the in-situ vertical effective
stress (σ v0 ); the undrained strength ratio lies in the range 0.2 < su /σv0 <
′ ′

caused by the included mud. In the case of Unit 4, it appears to be

disturbed/remolded Bay Mud; its properties have been taken as the 0.4, a usual range for normally consolidated clay. The undrained
weak end of the spectrum of Layer 5 discussed subsequently. strength ratio was also computed from the CPT using the cone factor
Nkt = 16 (a value routinely used for HBM); this data also being shown in
Fig. 9 but now plotted against the normalized excess pore pressure co­
4.3. Bay Mud (Layer 5)
efficient Bq since void ratio was unknown at each CPT data scan. The
range and mean strength ratio of strengths from the two CPT soundings,
Undisturbed samples of the Holocene Bay Mud were tested in the
ITP-3 and ITP-4, was similar to strengths measured with laboratory
laboratory. Both compressibility (oedometer) and the undrained
samples and largely also unaffected by Bq similar to the lack of effect of
strength (undrained triaxial tests) were measured. These undisturbed
void ratio seen in the compression test. Conversely, the other two CPT
samples showed high water contents, presented as a void ratio versus the
soundings, ITP-1 and ITP-2, while exhibiting similar strength ratios to
in-situ vertical effective stress of the sample (σv0 ) on Fig. 8 below; void
′

the laboratory compression tests for Bq > 0.7 (= truly normally

ratios were commonly greater than 2. The CSL and ψ contours are
consolidated as conventionally understood), show a more usual trend of
approximated to match the observations in Figs. 18 and 19 (discussed
improving strength with decreasing Bq (i.e., an apparent effect of in situ
later) and are shown only for illustrative purposes (see Fig. 8’s caption).
state).
The sand fill had been in place for at least 15 years when the first
The undrained strength of soils scales with Mtc . Meehan et al. [29]
oedometer samples were taken, and full dissipation of fill-induced excess
carried out drained direct shear tests on remolded Bay Mud and ob­
pore pressure should have occurred within about six years based on the
tained large-deformation values in the range 23.3 deg < φcv < 25.2 deg;
measured coefficients of consolidation (cv ).
taking an average of this range, this is equivalent to Mtc ~ 0.95. Sitar and
The HBM surface elevation indicates that the HBM was always sub­
Salgado [26] reported on the response of the Buy Mud in cyclic simple
merged by the Bay as the surface was below mean low water elevation
shear, their data being presented after normalization by the static un­
with desiccation unlikely. Thus, no apparent over-consolidation from
drained strength. A reasonable average strength is su /σv0 ~ 0.3 from the
′

aging or other geological processes would be expected. Other workers

laboratory tests shown in Fig. 9a, which then gives a cyclic strength
trend from the Sitar and Salgado [26] data, as shown in Fig. 10. This
Table 1
figure also shows several other examples for silts, and the trend for Bay
Summary of fill properties.
Mud is not exceptional.
Soil D50 (μm) Fines λ10 Mtc N X

HBBP (2020) 220 5% 0.09 1.24 – –

Isserk 220/5 220 5% 0.086 1.26 0.3 2.7
Isserk 220/10 220 10% 0.123 1.26 0.3 2.7

5
J. Macedo et al.
Fig. 9. Undrained strength ratio of Holocene Bay Mud, a) Undrained strength
of undisturbed samples from laboratory tests, b) undrained strength from CPT.
Fig. 10. Cyclic strength of Bay Mud in cyclic simple shear tests compared to
other silts.
5. Soil state
5.1. Overview of CPT processing
Soil state (density, over-consolidation, state parameter) is primarily
assessed from CPT data. The CPT measures a mechanical response that
depends on both the soil’s properties (e.g., compressibility as identified
by Campanella et al. [30]) and the in situ state parameter so that
assessing the soil state from the CPT should account for the soil’s
properties – a soil-specific calibration is needed. However, soil proper­
ties can vary somewhat within a stratum and certainly between strata. If
a stratum can be simply identified; for instance, the HBM unit is clear in
6
Soil Dynamics and Earthquake Engineering 158 (2022) 107280
Fig. 3, a representative calibration can be assigned for that formation
based on the measured or estimated soil properties. This calibration can
be computed for both drained and undrained penetration, for example,
using scaled cavity expansion theory (Shuttle and Jefferies [31]). If the
stratum contains either a mix of soils or is a transitional boundary be­
tween strata, then the effect of changing properties should be part of the
data processing; this is done using the notion that the CPT senses ‘soil
type’ (the familiar idea of soil classification charts for the CPT, Fig. 12
being an example of this) and soil types have ‘representative’ properties.
Thus, the data processing can be automated to capture the effect of
changing soil type using the time-synchronized CPT measurements, i.e.,
tip resistance (qc ), sleeve friction (fs ), and pore pressures (u2 ) triplets
(“scans,” which are commonly at 20 mm intervals). The first step in
doing this was the ‘Plewes Method’ (Plewes et al. [32]) which went even
further and unified both drained and undrained calibration by assuming
that including the measured u2 in the data processing was sufficient to
capture the difference between drained and undrained conditions.
The Plewes Method is widely used in practice, even in such signifi­
cant engineering as understanding the Brumadinho dam failure by static
liquefaction (Robertson et al. [14]). But, the Plewes Method was
developed using the limited data on soil response to the CPT available in
the early 1980s and has never been updated. Recent data suggests that
the embedded relation between the normalized CPT friction ratio (F)
and the CPT-based soil classification index remains valid on average but
with quite substantial soil to soil variability (Reid [33]). It is also now
known that there is a clear step-change in calibration at the drained to
undrained transition, which occurs at about IcBJ (the soil classification
index defined in Been and Jefferies [34]) ~ 2.3. The CPT processing used
here involves a modified Plewes Method for the sand fill units, where
penetration is drained with the range in gradation influencing the CPT
calibration; the modification is described in the next sections. The HBM
unit is processed using a single soil-specific calibration, as also discussed
later. The Merritt sand and deeper strata are assessed in a screening-level
assessment from available CPTs.
5.2. Comparison of CPT before and after the 1989 Loma Prieta
earthquake
Automated procedures like the Plewes Method require time-coherent
data, effectively only being usable with digital records. But, such digital
recording did not become standard until the late 1980s. Thus, a detailed
evaluation of this site depends on post-earthquake measurements from
1989 rather than the pre-earthquake 1984 CPT that only had paper re­
cords – and which leads to the question: did the earthquake affect the
CPT data? This question is considered first before characterizing the
strata identified in Fig. 3.
There are locations where CPTs were carried out during 1984, at
which time the fill was about ten years old, and where CPTs were pushed
at a nearby location after the earthquake. Example comparisons are
presented in Fig. 11, the comparisons being made by scanning the 1984
paper records and then overlaying the 1989 data. The pale grey lines on
Fig. 11 are 1984 data, the dark lines the matching 1989 post-earthquake
pair.
The middle-overlay of Fig. 11, the ITP-5 pair, exhibits a few depth
misalignments, but that is normal for CPT soundings even as close as a
few meters. The qc values of the loose sand fill and underlying HBM are
similar between 1984 and 1989 with possibly a small reduction of tip
resistance in the loose sand; thus, a small effect of the earthquake on CPT
resistance at this location within the precision of the 1984 record is a
possible slight reduction of tip resistance.
In the left-hand overlay of Fig. 11, the ITP-2 pair, the zone annotated
as A has much less resistance post-earthquake; plausibly, this could be
where there was an upward movement of sand after liquefaction
(‘ejecta’). Apart from this loosened zone, the remainder of the profile is
consistent with the 1989 data being an adequate representation of the
site prior to the earthquake.
J. Macedo et al. Soil Dynamics and Earthquake Engineering 158 (2022) 107280

Fig. 11. Comparison of paired CPT resistance profiles before and after the Loma Prieta earthquake.

The last overlay of Fig. 11, the ITP-7 pair, gives an opposite view
with clear densification of the zone annotated as B.
It is standard to use post-earthquake measurements as indicative of
the prior conditions (e.g., Mitchell et al. [35]), and that is forced for
detailed analysis of this case history because only CPT campaigns later
than 1989 had the needed digital data acquisition. The comparison of
this digital data with the older paper records suggests that, on average,
this is acceptable but that the 1989 records will have greater dispersion
of the tip resistance within the loose sand fill. As far as can be judged,
because of the limited available resolution of the 1984 data, the HBM
shows a similar response in 1989 as 1984.

5.3. Hydraulic sand fill

The results of the 1989 CPT soundings within the sand fill, layers 1
and 2 on Fig. 3, are shown in the Jefferies and Been [24] ‘soil behavior
type’ chart in Fig. 12. This chart also shows “screening” level contours of
ψ , which are useful for an initial appreciation. The upper 2 m or so,
Layer 1, is the ‘ubiquitous dense’ stratum across the site and is very
dilatant (ψ < − 0.2) and presents as clean sand. In parts, this layer is very
dense. It is above the water table, and thus its only role in the lique­
faction is as an overlying stiff mass. The underlying sand fill, Layer 2,
plots as rather heterogeneous with soil type ranging from ‘sand, some
silt’ through to ‘silty sand’; exactly as expected from the gradations
determined on the fill samples. The data tends to follow a constant-state
Fig. 12. CPT data within the hydraulic fill as a soil behavior type.
contour (ψ ≈ − 0.1), an aspect of soil development/accretion that has
been found elsewhere (Reid and Jefferies [36]); in essence, the soil
naturally compensates for the changes in gradation from place to place

7
J. Macedo et al. Soil Dynamics and Earthquake Engineering 158 (2022) 107280

within a similar depositional environment giving similar mechanical

behavior despite the changes in gradation.
The silt and clay inclusions of Layers 3 and 4 show no systematic
trend to parallel a constant-state contour; at this level of data compila­
tion, these layers show an almost random range of gradations and an
equally random range of state: from slightly contractive to very
contractive. Based on the understanding of Layers 3 and 4 from the
borehole records, these two layers were distinguished using the
measured Bq : Bq > 0.05 was taken as Layer 4. If not Layer 4, the fill was
processed using a drained inversion with the inversion coefficients
depending on F as detailed in the following.
The in-situ ψ depends on the stress-normalized CPT tip resistance Q
through:
m ψ = − ln(Q / k) [1]

where k, m are the soil-property-dependent inversion coefficients. These

coefficients were computed using Norsand (Jefferies [37]) in cavity
expansion, scaling to the worldwide database of CPT calibration
chamber studies (Ghafghazi and Shuttle [38]). The properties used are
those given in Table 1, with an example of the numerical results being Fig. 14. ITP-specific calibration for inversion coefficient k in sand fill.
shown in Fig. 13 for the 10% fines sand. There is a small effect of the
elastic rigidity (Ir ) on the CPT resistance, but this has been neglected by
using an average trend because it is second-order to the more important
effect of changing gradation.
The CPT data plotted in Fig. 12 suggests that the least-fines fill, with
about 5% fine contents (Fig. 4), can be associated with F ~ 0.3%. The
average fines content of the clean fill is 10%, and there is a clear con­
centration of data at F ~ 0.4%. The computed k for the properties given
in Table 1 have been associated with these respective values and plotted
in Fig. 14; the Plewes Method trend for an average sand with Mtc = 1.25
is also shown in this figure. The ITP-specific calibration for both 5% and
10% fines is weaker (i.e., lower k) than the backbone-trend in Plewes
Method; the dashed line shows the modification to the Plewes Method to
account for site-specific calibration.
The second inversion coefficient (m =5.5) was similar for both the
5% and 10% fines fill. An average value was used regardless of the CPT
friction ratio. The drained-undrained transition was established by
plotting the measured Bq versus the soil type index (IcBJ), Fig. 15. There
is a near “wall” at IcBJ ~ 2.3 for this site (see Fig. 15). Accordingly, the
data processing to infer state within the fill considers k from Fig. 14 and
m = 5.5 for IcBJ < 2.3. Fig. 15. Drained-undrained transition in CPT soundings at ITP.
The results of processing the CPT data within the sand fill (i.e., the
soils lying above the top surface of the HBM) are shown in Fig. 16. Aside Fig. 12, with the calibrated procedure, indicating -0.05 < ψ < +0.2 for
from the very dense sand (ψ < -0.2), the fill presents systematically most of the data. In terms of liquefaction, the loose fill is obviously a
looser than inferred by the ‘screening level’ guidance contours on contributor to the site behavior during the Loma Prieta earthquake.

5.4. Holocene Bay Mud

Fig. 13. Example of computed CPT inversion considering various elastic ri­
gidity (Ir ) and initial mean pressure (p) values.
The south side of the ITP is underlain HBM, and the results of the
1989 CPT soundings within the Bay Mud are shown in Fig. 17. This
stratum plots as a brittle, sensitive silt, rather different from the
perception of a plastic normally consolidated soft clay. There are sandy
stringers within the HBM, which present as loose, but that is most likely
an artifact of having soft silt above and below (no ‘thin layer’ correction
has been used).
The NorSand implementation of critical state theory was calibrated
to the Bay Mud, the calibration being shown to monotonic simple shear
data reported by Rau and Sitar [27]) in Fig. 18. Test MONO-8 was
chosen as that was one of the loosest tested with a void ratio e = 2.04 at
an effective vertical stress of 100 kPa, a void ratio and stress combina­
tion that is within the range of the in situ conditions at the site albeit at
the denser end of the range. The calibration adopted λ10 = Cc (not un­
reasonable for a fine-grained material and Cc having been measured, as
discussed earlier), Cs /Cc = 0.1 (characteristic of measured values),
Mtc = 0.95 (measured, discussed earlier), N = 0 (a ‘not unreasonable’

8
J. Macedo et al. Soil Dynamics and Earthquake Engineering 158 (2022) 107280

Fig. 17. CPT data within the Holocene Bay Mud as soil behavior type.

Fig. 16. Inverted ψ below water table within the fill at ITP from 1989
CPT soundings.

value for silts) and over-consolidation ratio OCR = 1 (as reported by Rau Fig. 18. Calibration of NorSand (red) to Bay Mud Test MONO- 8 (blue) in
and Sitar [27]). Other properties, and the state parameter, were then monotonic simple shear tests. The MONO tests come from Rau and Sitar [27].
iterated to get a representative fit of the measured stress-strain curves;
the achieved fit was for a lightly dilatant state ψ = − 0.05, a
Jefferies [31], with an updated version of the software for direct use
state-dilatancy coefficient X = 2, and a plastic hardening modulus H =
with u2 data; this update was validated in the Cadia silt (see Figure E4-5
6.
of Morgenstern et al. [13]). The computed inversion coefficients, using
The calibrated soil properties were then carried forward into the
the HBM properties as per Fig. 18, were k’ = 6.5, m’ = 2.6, and the
inversion of the CPT data to compute the in situ state. The methodology
results of processing the CPT data with these coefficients are shown in
uses the measured CPT data, which is normalized using the excess pore
Fig. 19.
pressure at the u2 location:
A first point to note in Fig. 19 is the state used to calibrate test DSS-
′
m ψ = − ln(Q / k )
′
[2] MONO8 (Fig. 18), which is well-aligned with the denser computed in-
situ state and consistent with that test being at the low end of the in-
where situ void ratios. The in-situ state range computed from the CPT is also
( ) consistent with the data in Fig. 8. Going further, the “classical geolog­
Q = Q 1 − Bq + 1 [3] ical” idealization of ‘normal consolidation’ is for the NCL to parallel the
CSL in e − log(p’) space with an offset of about a factor of 2 measured on
and k’, m’ are the soil-property-dependent inversion coefficients. These the p-axis; if this is converted to void-ratio, ‘normal consolidation’
coefficients were computed using the cavity inversion in Shuttle and amounts to ψ ~ 0.8/(λ − κ), where λ is the CSL slope in natural log units

9
J. Macedo et al.
Fig. 19. Computed in-situ state of Bay Mud.
and κ the swelling index, or ψ ~ +0.3 in the case of the Bay Mud. As can
be seen in Fig. 19, the computed in-situ state from the CPT inversion is
generally a little denser than ψ = 0.3, consistent with the Bay Mud being
normally consolidated but then experiencing secondary consolidation
(aging). Thus, what is computed for the in-situ state is consistent with
the current understanding of the Bay Mud “in general.” But, where the
HBM at this site departs from accepted understanding is a layer between
7 m and 8 m depth present at ITP-2 and ITP-4 where the HBM is
markedly looser with ψ > +0.4; at the simplest level, this assessed
looseness is also evident from the CPT data as Bq > 0.6 in these
soundings (Fig. 9b).
The effect of this range of in-situ states on undrained strengths is
shown in Fig. 20, computed using the soil’s properties as discussed
above. The strengths at 15% strain are consistent with the undrained
strength data shown in Fig. 9, with the range in measured strengths
simply reflecting the range of state of the tested samples. The very weak
strength computed, and not seen in the test data, is likely because such
samples would have been classified as “disturbed” and thus not tested.
Although the Bay Mud presents on the ‘soil behavior type’ chart as
predominantly sensitive clayey silt, the computed stress-strain behavior
of the very loosest parts of the Bay Mud is simply just weak with minimal
post-peak strength loss (Fig. 20). This computed response is a conse­
quence of the soil’s extraordinarily large compressibility.
Fig. 20. Computed undrained triaxial compression of Bay Mud using Norsand
calibrated properties.
10
Soil Dynamics and Earthquake Engineering 158 (2022) 107280
5.5. Older sediments
Data for the older sediments come from more recent (2018) CPT
soundings in an adjacent area of the HBBP, where three soundings were
pushed to some 25 m depth. The results are shown as soil behavior type
in Fig. 21.
The Merritt Sand presents as dense sand, some silt grading to silty
sand in parts. The soundings in this unit showed negative excess pore
pressure (see Fig. 3), indicating at least only partial drainage; this sug­
gests this unit is gradationally finer than indicated by its soil behavior
type. The density of these sands suggests a transgressive marine depo­
sition with wave-induced densification; at a ‘screening level’, the Merritt
Sand exhibits − 0.10 > ψ > − 0.20, although this range reflects system­
atic change across the site with CPT-2018-11A encountering looser
conditions (ψ k ~− 0.12) than CPT-2018-10A (ψ k ~− 0.17) throughout
the Merritt and with the third CPT-2018-09A lying in the middle.
As previously discussed, there are multiple sub-units within the un­
derlying Pleistocene Bay Mud. Layer 7 presents as over-consolidated
and/or dense to the extent of being quite dilatant; the undrained
strength ratios commonly exceed su /σv0 > 0.7 using Nkt = 16. Layer 8,
′
despite being identified as a separate layer on the depth record, is pre­
sent in the same place in terms of soil behavior type. Layer 9, despite
underlying 7, appears as slightly more plastic but certainly less over-
consolidated; this could reflect a lower friction angle and/or greater
compressibility rather than the geologically difficult less over-
consolidation per se. Undrained strength ratios are largely in the
range 0.4 < su /σ v0 < 0.6 using Nkt = 16.
′
6. Ground motion during event
The October 17, 1989 Loma Prieta earthquake had a magnitude of
6.9, with an epicenter at approximately 16 km northeast of Santa Cruz.
The effects of the Loma Prieta earthquake extended well to the north
into the San Francisco Bay area, including Alameda (at an epicentral
distance of some 85 km). The recorded ground motion at the Yerba
Buena Island (YBI; VS30 = 660m/s, where VS30 is the time-averaged
shear-wave velocity in the top 30 m) has been used as that of bedrock
for studies at Alameda, including the HBBP. The nearest site to HBBP
Fig. 21. CPT data within the Pleistocene sediments, plotted by soil
behavior type.
J. Macedo et al.
with a recording strong-motion station at the time was the Naval Air
Station (NAS; VS30 ≈ 150 m/s), about 7 km distant, and the outcrop
motion at YBI correlates with that of the surface at the NAS with an
approximate time delay of 320 ms, illustrated in Fig. 22. Of note, the
same stratigraphic sequence is found at HBBP and NAS; in addition, the
VS30 are similar for the two sites, thus we will use it as the surface ground
motion. The motion of the NAS site stays aligned with that of the
bedrock until approximately 12 seconds elapsed time (taking the origin
of the NAS record as the start of the recording). At that instant, the NAS
site can no longer follow the motion of the bedrock and the two motions
decouple. It is inferred this decoupling is caused by liquefaction or
similar of the upper loose soils at 12 seconds, with the upper part of the
site (including the hangar with the strong motion station) then
“wobbling” on the softened/liquefied ground largely independent of
basal acceleration. The ratio of accelerations at the top and bottom parts
of the soil profile is approximately constant (a factor of 3.75) during the
time interval 0–12 seconds of the two records, as observed in Fig. 22,
where the 3.75 factor is already applied. The decoupling can also be
observed by assessing the change in the frequency content of the ground
motion through the use of a Stockwell transform (Stockwell et al. [39]),
which depicts the time-frequency distribution of a ground motion
recording. Fig. 23 shows the Stockwell transforms for the NAS and Yerba
Buena records; it can be observed that the large frequencies (i.e., larger
than 5hz) in the Yerba Buena record are not present in the NAS record
after approximately 12 s when the most intense part of the ground
motion occurs which may have caused increased excess pore pressures
and a loss in stiffness, which in turn shifts the motion to higher periods
(lower frequencies). Similar observations were made for ground motions
recorded at liquefied sites by Kramer et al. [40] and Macedo and Bray
[41].
We also focus on the ground motion between recorded times of 8
s–16 s, which is the portion of the record where the strongest motion
occurred. We estimated that there are about 5 ‘significant’ cycles within
this portion of the record which were subsequently followed by much-
reduced amplitude motions for a further 10 seconds or so. The num­
ber of ‘significant’ loading cycles is about half what would usually be
associated with an earthquake of Loma Prieta’s magnitude, and it may
be related to the particularities associated with the fault rupture in the
Loma Prieta earthquake.
7. Liquefaction observations
7.1. Manifestation of liquefaction
Liquefaction at the HBBP was manifested as sand boils south of
Harbor Bay Parkway, in the ITP, west of South Loop Road in the Airport
Fig. 22. Comparison of ground motions at the NAS recording station with the
bedrock ground motion recorded at the YBI station.
11
Soil Dynamics and Earthquake Engineering 158 (2022) 107280
Center and surgical dynamics sites, and along the north side of Harbor
Bay Parkway north and west of the ITP. Sand boils were apparent around
the ITP buildings and in the undeveloped areas throughout the 0.12 km2
ITP site, see Fig. 1. There were several locations in the HBBP where thin
cracks were apparent in the ground surface. Some of these cracks
extended for substantial distances and were side by side and acted as
conduits for the upward movement of sand and water that emerged as
sand boils. Other man-made conduits, such as the location of manholes,
drop inlets, and risers for pipes, provided similar conduits. Typically, the
pressurized sand and water flowed up along the side of these structures
to the ground surface and migrated out onto the ground surface; Fig. 24
below illustrates this aspect.
Two identical two-story concrete-framed office buildings had been
built within the ITP before 1989. It was recognized that the loose fill
sands could liquefy, and the buildings were designed with this in mind.
The two structures were supported on spread footings bearing in the
ubiquitous dense sand that caps the site, with footing elevations
providing a distance of at least twice footing width to the base of the
dense sand. Numerous sand boils were apparent around the ITP build­
ings after the earthquake. Ejected sand emerged through a saw cut made
through the ground floor slab in one building a few days before the
earthquake, filling the room with a few inches of sand. Earthquake-
induced settlements (estimated from notes made at the time) were in
the order of 100–150 mm, some of which may have been caused by “loss
of ground” associated with the development of sand boils at the ITP.
8. Site liquefiability
8.1. Cyclic stress ratios
The shear stresses at a given depth were estimated as τ = agσv rd where
a is the acceleration time history at the surface, σv is the total stress at a
given depth, and rd is a depth-dependent factor that modifies the a to
account for the flexibility in the dynamic response of a soil mass. We
used the rd factors from Boulanger and Idriss [42]. The estimated τ
time-history at the middle of layer 5 (~7.5 m depth) is presented in
Fig. 25. The cyclic stress ratio (CSR) at a given depth was estimated as
CSR = 0.65 τσmax′ , where σv is the effective vertical stress at a given depth.
′
v
For example, CSR for layer 2 varies from 0.10 to 0.15.
8.2. Holocene Bay Mud
The cyclic strength of the Bay Mud at five cycles, which is the esti­
mated number of significant cycles in the Loma Prieta record, derives on
a cyclic resistance ratio (CRR) of about 0.25 (Fig. 10). This is less than
the peak demand apparent in Fig. 25. Focusing on strength misses an
important aspect of Bay Mud behavior. Referring to Fig. 20, the Bay
Mud’s stress-strain behavior, regardless of ψ , is near bi-linear with a
relatively stiff initial response followed by an important loss of stiffness
once the mobilized strength ratio sy /σ v0 > 0.2 (and possibly as little as
′
0.16) where we are now associating the loss of stiffness as a “yield”
condition. The strength-demand represented by the CSR time history
shown in Fig. 25 could not be transmitted by the Bay Mud beyond the
second-pulse at about 11 s elapsed time when τ > sy , which is exactly the
behavior seen in the strong-motion record at the geologically similar
NAS site and at near-enough the same point in the record (see Fig. 22 at
12 seconds). It is not that the Bay Mud is ‘liquefiable’ but rather that Bay
Mud is simply a weak clayey silty material with an abrupt yield into a
near fully-plastic deformation; hence, limiting the stresses transmitted
to the overlying sands (see discussion section).
8.3. Sand fill
Liquefaction assessment charts/procedures, and strength trends, are
normalized to 15 loading cycles. As there were approximately 5 cycles in
J. Macedo et al. Soil Dynamics and Earthquake Engineering 158 (2022) 107280

Fig. 23. Top: Stockwell transform of the acceleration-time history recorded at the top of the NAS station. Bottom: Stockwell transform of the acceleration-time
history recorded at the Yerba Buena station.

Fig. 24. Example of liquefied sand transported to the surface of the building
periphery (1320 Harbor Bay Parkway).
Fig. 25. Estimated cyclic stress ratio history at the middle of Layer 5.
the Loma Prieta record, the sand fill at the ITP will behave ‘stronger’
because of the reduced demand on it. Laboratory data suggest that a 15- the transmitted shear stresses, presents as liquefiable.
to-5 cycle scaling would amount to an apparent strength increase by a
factor of approximately 1.3–1.4 (Jefferies and Been [43]).
8.4. Merritt Sand
The in-situ state parameter of the majority of the loose fill, Layer 2 of
the site characterization (Fig. 3), shows natural variability. Stochastic
In terms of the liquefaction response of the Merritt sand during the
simulations (Popescu et al. [44–47]) indicate that about the loosest 20%
Loma Prieta earthquake, a comparison of the assessed characteristic
of a formation controls its behavior, commonly referred to as the char­
state of the Merritt sand (− 0.12 > ψ k > − 0.17) discussed earlier with the
acteristic value from the distribution (and denoted by the subscript ‘k’).
liquefaction triggering criteria shown in Fig. 26, derives in a CRR of
A reasonable judgement is that layer 2 has -0.02 < ψ k < 0.10.
about 0.25, which when compared against the estimated CSR at the top
The assessed demand and in-situ states are shown on a liquefaction
of the Merritt sand, which is in the order of 0.16 using the procedures in
triggering chart, Fig. 26. This chart uses the “Class A” case-histories of
the “cyclic stress ratio” section (after a correction for the number of
Moss [48], updated to a characteristic state parameter (Jefferies and
cycles as discussed in section 8.1), shows the Merritt sand as not lique­
Been [43]), which is only valid for F < 1.3. Most, but not all, of the data
fiable. A further factor is that the Merritt sand presents on the soil
for Layer 2 fall within this friction ratio limit of validity (Fig. 12). The
behavior type chart (Fig. 21) as relative compressible, a factor that will
loose hydraulic fill, despite the potential role of the Bay Mud on limiting
also attenuate induced excess pore pressures during cyclic loading. Thus,

12
J. Macedo et al.
Fig. 26. Comparison of assessed in-situ state with vulnerability to earthquake-
induced liquefaction. Note that the x-axis is inverted to highlight negative
state parameters.
it is reasonable to treat the Merritt sand as having maintained its shear
stiffness during the Loma Prieta earthquake.
9. Discussion
The classification of fine-grained soils as “clay-like” is purely
geological and neglects their intrinsic behavior. On a geological classi­
fication, the Bay Mud would be “clay-like” as per Idriss and Boulanger
[49] (the plasticity index is significantly higher than 7) and treated for
liquefaction purposes accordingly. A feature of “clay-like” is that only
stress-history is important; however, the Bay Mud shows a spectrum of
void ratios with no normal compression zone apparent when plotting
in-situ void ratio versus effective stress of the sample (Fig. 8). In fact,
regarding the field void ratios measured in the Bay Mud, if the y-axis
scale were hidden from the viewer, the range of void ratios would look
as “sand-like.” This is not a limitation for the selected approach in this
study as critical state theory does not allow separation into “sand-like”
or “clay-like” behavior, with the same equations applying across the
spectrum from soft clay through to (at least) coarse sand – it is just a
question of the numerical values for the soil properties. Interestingly, the
application of the critical state theory to Bay Mud gives a consistent fit of
the computed state and the corresponding undrained strength to inde­
pendent measurements. This, in turn, then shows that Bay Mud is a weak
soil with full-plastic yielding at about half the demand of the Loma
Prieta earthquake. Thus, focusing on a critical layer within the sand
simply misses the importance of the Bay Mud and its interaction with the
fill units. In this regard, looking to the full profile of the soil on the CPT
leads to the same conclusions as Cubrinovski [50]: site details matter.
In terms of the interaction between the Bay Mud and the fill layers,
strictly speaking, the monotonic response of the Bay Mud (presented in
Fig. 20) does not answer the issue of unloading between cycles; how­
ever, theory suggests that unloading-reloading is stiff even when the
original loading may have reduced the soil’s strength significantly.
Moreover, our computed stress-strain behavior allowing for the looser
in-situ states does not predict strain-softening (see Fig. 20). Thus, this
suggests that the role of the Bay Mud is limiting the transmitted shear
stresses rather than acting as a true filter of high-frequency motion.
Hence, it is reasonable to expect that much of the filtering may have
occurred on the liquefied fill, with some contribution from the Bay Mud,
especially during the strong part of the shaking. This compounded
behavior for the Bay Mud and the loose fill is consistent with the sudden
decoupling of bedrock and site motions at the geologically similar Naval
Air Station. Finally, despite the potential role of the Bay Mud in limiting
13
Soil Dynamics and Earthquake Engineering 158 (2022) 107280
the transmitted stresses, the evidence of sand boils (Fig. 24) suggests
that this was not sufficient to protect the hydraulic sand fill: there was
sand liquefaction at the ITP. This should not come as a surprise as, even
after scaling for the short duration of the Loma Prieta event, the in-situ
state of the fill sands was such that the onset of earthquake-induced
liquefaction was to be expected.
10. Conclusion
The vulnerability of the International Telegraph Plaza (ITP) to
earthquake-induced liquefaction has been assessed using a critical state
approach; this approach explicitly includes soil properties as opposed to
only fine contents corrections, which are not fundamentally anchored on
mechanics in our view. A consistent pattern of the various measure­
ments is revealed, with the inferred state of the hydraulic fill and the soft
underlying Bay Mud, leading to an expected performance in the Loma
Prieta earthquake that matches that encountered. Interesting aspects of
this approach were the ease with which actual soil properties were
included and the further ease of dealing with the wide range of soil
gradations within the fill.
Author statement
Jorge Macedo: Conceptualization, Writing- Original draft, Meth­
odology, Investigation, Supervision, Funding acquisition. Luis Ver­
garay: Data curation, investigation, Writing- Original draft. Curtis
Jensen: Visualization, Investigation, Resources. Renzo Cornejo: Visu­
alization, Writing- Original draft. Michael Jefferies: Conceptualization,
Writing- Original draft, Data Curation.
Declaration of competing interest
The authors declare that they have no known competing financial
interests or personal relationships that could have appeared to influence
the work reported in this paper.
Acknowledgment
We acknowledge the financial support provided by the Civil and
Environmental Engineering department of the Georgia Institute of
Technology and the financial support provided by the Pronabec program
of the Peruvian government for the second and fourth authors. The
authors also appreciated Langden’s sharing of their deeper CPT data for
the area to supplement the extensive but shallow data from the author’s
own work. We also thank Mr. Virgil Baker, who provided post Loma
Prieta earthquake digital CPT records on the ITP area. Finally we also
thank prof. Nicholas Sitar for his insightful comments on the nature of
HBM and its influence on the ground motion at the NAS site.
References
[1] Mitchell JK, Lodge AL, Coutinho RQ, Kayen RE, Seed RB, Nishio S, Stokoe KH. In
situ test results from four Loma Prieta earthquake liquefaction sites: SPT, CPT,
DMT and Shear Wave Velocity: Berkeley. Report UCB/EERC-94/04. University of
California Earthquake Engineering Research Center; 1994. p. 179.
[2] Moss RE, Seed RB, Kayen RE, Stewart JP, Der Kiureghian A, Cetin KO. CPT-based
probabilistic and deterministic assessment of in situ seismic soil liquefaction
potential. J Geotech Geoenviron Eng 2006;132(8):1032–51.
[3] Hallenbeck, Associates. Response of ground to Santa Cruz earthquake (sic) of
October 17, 1989. Letter to Doric Development; October 20, 1989.
[4] Youd TL, Idriss IM, Andrus RD, Arango I, Castro G, Christian JT, Dobry R, Liam
Finn WD, Harder Jr LF, Hynes ME, Ishihara K, Koester JP, Liao SSC,
Marcuson III WF, Martin GF, Mitchell JK, Moriwaki Y, Power MS, Robertson PK,
Seed RB, Stokoe II KB. Liquefaction resistance of soils: summary report from the
1996 NCEER and 1998 NCEER/NSF workshops on evaluation of liquefaction
resistance of soils (NCR method). J Geotech Geoenviron Eng 2001;127:10. https://
doi.org/10.1061/(ASCE)1090-0241(2001)127:10(817).
[5] Lyman. Construction of Franklin falls dam. Report, US Army Corps of Engineers.
1938.
J. Macedo et al.
[6] Casagrande A. Characteristics of cohesionless soils affecting the stability of slopes
and earth fills. Harvard University; 1936.
[7] Casagrande A. Liquefaction and cyclic deformation of sands, a critical review.
Buenos Aires: Proceedings of 5th Pan American Conference Soil Mechanic,
Foundation Engineering; 1975. p. 80–133.
[8] Taylor DW. Fundamentals of soil mechanics. New York: John Wiley; 1948.
[9] Bishop AW. Reply to discussion on “Measurement of shear strength of soils” by A.
W. Skempton and A.W. Bishop. Géotechnique 1950;2:90–108.
[10] Been K, Jefferies MG. A state parameter for sands. Geotechnique 1985;35(2):
99–112.
[11] Jefferies MG. On the fundamental nature of the state parameter. Géotechnique
2021. https://doi.org/10.1680/jgeot.20.P.228.
[12] Morgenstern N, Vick S, Viotti C, Watts B. Fundão tailings dam review panel report
on the immediate causes of the failure of the Fundão dam. Samarco S.A., Vale S.A;
2016.
[13] Morgenstern NR, Jefferies M, Zyl DV, Wates J. Independent technical review
board. Report on NTSF embankment failure. Ashurst Australia; 2019. Available at:
https://www.newcrest.com/sites/ default/files/2019-10/190417_Report%20on%
20NTSF%20Embankment%20Failure%20at%20Cadia %20for%20Ashurst.pdf.
[14] Robertson PK, De Melo L, Williams DJ, Wilson GW. Report of the expert panel on
the technical causes of the failure of Feijão dam I. 2019. Available at: http://www.
b1technicalinvestigation.com/.
[15] Arroyo M, Gens A. Computational analyses of dam I failure at the Corrego de Feijao
mine in Brumadinho – final report. Vale S.A; 2021.
[16] Stamatopoulos CA. An experimental study of the liquefaction strength of silty sands
in terms of the state parameter. Soil Dynam Earthq Eng 2010;30(8):662–78.
[17] López-Querol S, Coop MR. Drained cyclic behaviour of loose Dogs Bay sand.
Geotechnique 2012;62(4):281–9.
[18] Qadimi A, Mohammadi A. Evaluation of state indices in predicting the cyclic and
monotonic strength of sands with different fines contents. Soil Dynam Earthq Eng
2014;66:443–58.
[19] Pan K, Yang ZX. Effects of initial static shear on cyclic resistance and pore pressure
generation of saturated sand. Acta Geotechn 2018;13(2):473–87.
[20] Wei X, Yang J. Cyclic behavior and liquefaction resistance of silty sands with
presence of initial static shear stress. Soil Dynam Earthq Eng 2019;122:274–89.
[21] Rogers JD, Figuers SH. Engineering geologic site characterization of the greater
Oakland Alameda area, Alameda and san Francisco Counties, California. Report to
the National Science Foundation, report number: Grant No. BCS 9003785. 1992.
https://doi.org/10.13140/RG.2.1.4785.6809.
[22] Langan Engineering and Environmental Services, Inc. Geotechnical investigation
Harbor Bay South Loop. Report to srmErnst Development Partners. Alameda
California; 29 March 2019.
[23] USGS. U.S. Geological Survey. 1992. https://pubs.er.usgs.gov/publication
/70043734.
[24] Jefferies M, Been K. Soil variability and characteristic states. Soil liquefaction: a
critical state approach. CRC Press; 2016. p. 203–24.
[25] Jefferies MG, Been K. Soil liquefaction – a critical state approach. London and New
York: Taylor & Francis Group; 2006, ISBN 0-419-16170-8.
[26] Sitar N, Salgado R. Behavior of the san Francisco Bay Mud from the Marina district
in static and cyclic simple shear. Geotechnical Engineering Report No. UCB/GT/
93-03. Berkeley, CA: Department of Civil Engineering, University of California;
April 1993. p. 20.
[27] Rau GA, Sitar N. Post-cyclic response of Holocene Bay Mud from san Francisco.
Geotechn Earthq Eng Soil Dynam 1998;III:246–57.
[28] Parks MC. Engineering properties and geologic setting of old Bay clay deposits.
Downtown San Francisco, California: UC Berkeley; 2019. ProQuest ID: Parks_
berkeley_0028E_19138. Merrit ID: ark:/13030/m5jx3mp6. Retrieved from, htt
ps://escholarship.org/uc/item/7zx826gw.
14
Soil Dynamics and Earthquake Engineering 158 (2022) 107280
[29] Meehan CL, Duncan JM, Brandon TL, Boulanger RW. An experimental study of the
dynamic behavior of slickensided surfaces. Report of a study performed by the
Virginia. Tech Center for Geotechnical Practice and Research; 2006.
[30] Campanella RG, Robertson PK, Gillespie D. Seismic cone penetration test. Use of in-
situ tests in geotechnical engineering (GSP 6). Reston/VA: ASCE; 1986. p. 116–30.
[31] Shuttle D, Jefferies M. Determining silt state from CPTu. Geotechn Res 2016;3(3):
90–118. https://doi.org/10.1680/jgere.16.00008. Available at:.
[32] Plewes HD, Davies MP, Jefferies MG. CPT based screening procedure for evaluating
liquefaction susceptibility. In: Proceedings of the 45th Canadian geotechnical
Conference; 1992. Canada, Toronto.
[33] Reid D. Estimating slope of critical state line from cone penetration test—an
update. Can Geotech J 2015;52(1):46–57. https://doi.org/10.1139/cgj-2014-
0068.
[34] Been K, Jefferies MG. Towards systematic CPT interpretation. In: Proceedings of
the Wroth memorial symposium. London: Thomas Telford; 1992. p. 121–34.
[35] Mitchell JK, Wentz Jr FJ. Performance of improved ground during the loma Prieta
earthquake. University of California Berkeley, Earthquake Engineering Research
Center; 1992. Report No. UCB/EERC-91/12.
[36] Reid D, Jefferies M. State parameter as a geological principle in tailings. In:
Wilson G Ward, Sego DC, Beier NA, editors. Proceedings of tailings and mine waste
2017. Edmonton: University of Alberta Geotechnical Center; 2017. p. 305–14.
[37] Jefferies MG. Nor-Sand: a simple critical state model for sand. Geotechnique 1993;
43(1):91–103. https://doi.org/10.1680/geot.1993.43.1.91\. Available at:.
[38] Ghafghazi M, Shuttle D. Interpretation of sand state from cone penetration
resistance. Geotechnique 2008;58(8):623–34. https://doi.org/10.1680/
geot.2008.58.8.623.
[39] Stockwell R, Mansinha L, Lowe RP. Localization of the complex spectrum: the S
transform. IEEE Trans Signal Process 1996;44(4):998–1001. https://doi.org/
10.1109/78.492555.
[40] Kramer S, Sideras S, Greenfield M. The timing of liquefaction and its utility in
liquefaction hazard evaluation. Soil Dynam Earthq Eng 2016;91:133–46. https://
doi.org/10.1016/j.soildyn.2016.07.025.
[41] Macedo J, Bray J. Key trends in assessment of Liquefaction-Induced building
settlement from numerical analyses. Am Soc Civ Eng (ASCE/JGGE) J 2018;144
(11). https://doi.org/10.1061/(ASCE)GT.1943-5606.0001951.
[42] Boulanger R, Idriss IM. CPT-Based liquefaction triggering procedure. J Geotech
Geoenviron Eng 2016;142(2).
[43] Jefferies MG, Been K. Soil liquefaction: a critical state approach. second ed. Boca
Raton, FL, USA: CRC Press; 2015. https://doi.org/10.1201/b19114. Available at:.
[44] Popescu R, Prevost JH, Deodatis G. Effects of spatial variability on soil liquefaction:
some design recommendations. Geotechnique 1997;475:1019–36.
[45] Popescu R, Prevost JH, Deodatis G. Spatial variability of soil properties: two case
studies. Geotechnical earthquake engineering and soil dynamics. Geotechnical
Special Publication No. 75. Seattle: ASCE; 1998. p. 568–79.
[46] Popescu R, Deodatis G, Prevost JH. Simulation of non-Gaussian homogeneous
stochastic vector fields. Probabilist Eng Mech 1998;13(1):1–13.
[47] Popescu R, Prevost JH, Deodatis G. Characteristic percentile of soil strength for
dynamic analyses. Geotechnical Earthquake Engineering and Soil Dynamics.
Geotechnical Special Publication No. 75. ASCE; 1988. p. 1461–71.
[48] Moss RES. CPT-based probabilistic assessment of seismic soil liquefaction
initiation. Ph.D. dissertation. University of California Berkeley; 2003.
[49] Idriss IM, Boulanger RW. Soil liquefaction during earthquake. Oakland, California,
USA: EERI publication, Monograph MNO-12, Earthquake Engineering Research
Institute; 2008.
[50] Cubrinovski M. Some important considerations in the engineering assessment of
soil liquefaction. In: Acosta-Martinez H, Lehane BM, editors. Proc., 13th Australia
New Zealand Conf. on geomechanics; 2019. p. 19–36.